In the last decade, the space-borne experiment AMS-02 has determined cosmic-ray spectra with unprecedented precision, potentially providing
new insights into cosmic-ray propagation in our Galaxy. However, the analysis of this increasingly precise cosmic-ray data
requires more careful modeling of systematic uncertainties.
We analyze the recent AMS-02 comic-ray measurements of the secondaries Lithium, Beryllium, and Boron, and the primaries Carbon, Nitrogen and Oxygen.
The data is used to study five different propagation frameworks, which for example differ by including or neglecting reacceleration or breaks
in the diffusion coefficient. In this work, we particularly focus on systematic uncertainties related to the fragmentation cross section to
produce secondary CRs. Furthermore, the impact of correlated systematic uncertainties in the AMS-02 data are studied.
The cosmic-ray data from Li to O is well fitted in all the considered propagation frameworks. However, we find that the uncertainties on the nuclear production cross sections
play a crucial role. Because of these uncertainties we cannot distinguish between different propagation frameworks. In this sense, cross section uncertainties currently prevent
a deeper understanding of the properties of CR propagation.
Nonetheless, we find some common and robust constraints. Above a few GV, the slope of the diffusion coefficient is determined to be in the range of $\delta\simeq0.4-0.5$.
Moreover, we find a lower bound on the half-height of the diffusion halo at $z_\mathrm{h}\gtrsim3\;\mathrm{kpc}$.